Gas–solid fluidized bed reactors play an important role in many industrial applications. Nevertheless, there is a lack of knowledge of the processes occurring inside the bed, which impedes proper design and upscaling. In this work, numerical approaches in the Eulerian and the Lagrangian framework are compared and applied in order to investigate internal fluidized bed phenomena. The considered system uses steam/air/nitrogen as fluidization gas, entering the three-dimensional geometry through a Tuyere nozzle distributor, and calcium oxide/corundum/calcium carbonate as solid bed material. In the two-fluid model (TFM) and the multifluid model (MFM), both gas and powder are modeled as Eulerian phases. The size distribution of the particles is approximated by one or more granular phases with corresponding mean diameters and a sphericity factor accounting for their nonspherical shape. The solid–solid and fluid–solid interactions are considered by incorporating the kinetic theory of granular flow (KTGF) and a drag model, which is modified by the aforementioned sphericity factor. The dense discrete phase model (DDPM) can be interpreted as a hybrid model, where the interactions are also modeled using the KTGF; however, the particles are clustered to parcels and tracked in a Lagrangian way, resulting in a more accurate and computational affordable resolution of the size distribution. In the computational fluid dynamics–discrete element method (CFD–DEM) approach, particle collisions are calculated using the DEM. Thereby, more detailed interparticulate phenomena (e.g., cohesion) can be assessed. The three approaches (TFM, DDPM, CFD–DEM) are evaluated in terms of grid- and time-independency as well as computational demand. The TFM and CFD–DEM models show qualitative accordance and are therefore applied for further investigations. The MFM (as a variation of the TFM) is applied in order to simulate hydrodynamics and heat transfer to immersed objects in a small-scale experimental test rig because the MFM can handle the required small computational cells. Corundum is used as a nearly monodisperse powder, being more suitable for Eulerian models, and air is used as fluidization gas. Simulation results are compared to experimental data in order to validate the approach. The CFD–DEM model is applied in order to predict mixing behavior and cohesion effects of a polydisperse calcium carbonate powder in a larger scale energy storage reactor.